纯纯的ggplot2画好看的柱状图，统计、分面

1626501985614.jpg

如果你有一个表，有三列，一列数据，一列分组，一列组别

bar <- read.csv("bar.csv")
library(ggplot2) #画图
library(ggsignif) ### 统计

ggplot(bar,
       aes(x=Group,y=Value,color=Group,fill=Group))+
    geom_bar(stat="summary",fun=mean,position="dodge")+ #柱状图
    stat_summary(fun.data = 'mean_sd', geom = "errorbar", width = 0.5,position = position_dodge(0.9))+ ##'mean_sd' 自动计算均数+标准差，添加误差棒,当然也可以计算mean+se,mean_ci等，跟ggpubr一模一样，width可以设置误差棒的宽度，而0.9是误差棒的位置
    facet_grid(~Team,scales = 'free')+ #分面
    theme_minimal(base_size = 13)+ #主题和字体大小
    scale_color_manual(values = c('steelblue','firebrick'))+
    scale_fill_manual(values = c('steelblue','firebrick'))+
    geom_signif(comparisons = list(c("Control","Treat")),test = 't.test')+
    labs(x=NULL,y='Relative gene expression')

image.png

当然也可以在柱状图上添加点，这样的好处是可以看到原始数据

ggplot(bar,
       aes(x=Group,y=Value,color=Group,fill=Group))+
    geom_bar(stat="summary",fun=mean,position="dodge")+
    stat_summary(fun.data = 'mean_sd', geom = "errorbar", width = 0.5,position = position_dodge(0.9))+
    facet_grid(~Team,scales = 'free')+
    theme_minimal(base_size = 13)+
    scale_color_manual(values = c('steelblue','firebrick'))+
    scale_fill_manual(values = c('steelblue','firebrick'))+
    geom_signif(comparisons = list(c("Control","Treat")),test = 't.test')+
    labs(x=NULL,y='Relative gene expression')+
    geom_dotplot(stackdir = "center", binaxis = "y", 
                 fill = "lightgray", 
                 dotsize = 0.9,position = position_dodge(0.9)) #position很重要

image.png
如果不想显示具体的P值，还可以自动标星号， geom_signif里面加一句map_signif_level=T

ggplot(bar,
       aes(x=Group,y=Value,color=Group,fill=Group))+
    geom_bar(stat="summary",fun=mean,position="dodge")+
    stat_summary(fun.data = 'mean_sd', geom = "errorbar", width = 0.5,position = position_dodge(0.9))+
    facet_grid(~Team,scales = 'free')+
    theme_minimal(base_size = 13)+
    scale_color_manual(values = c('steelblue','firebrick'))+
    scale_fill_manual(values = c('steelblue','firebrick'))+
    geom_signif(comparisons = list(c("Control","Treat")),map_signif_level=T,test = 't.test')+
    labs(x=NULL,y='Relative gene expression')+
    geom_dotplot(stackdir = "center", binaxis = "y", 
                 fill = "lightgray", 
                 dotsize = 0.9,position = position_dodge(0.9))

image.png

当然最简单的是ggpubr，但是底层限制死了，很多细节不能DIY，比如误差棒的宽度

library(ggpubr)
ggbarplot(bar,'Group','Value',fill = 'Group',color = 'Group',
          facet.by = 'Team',scale='free',
          palette = c('steelblue','firebrick'),
          add = c('mean_sd','dotplot'),
          ggtheme = theme_minimal())+
    stat_compare_means(comparisons = list(c("Control","Treat")),label = "p.signif", method = 't.test')

image.png